INDEX

Explanations

sponsor or patron

The main thing this neuron does is detect mentions of sponsorship (forms of “sponsor/sponsors/sponsored/sponsoring”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 protože

-1.55

"'

-1.49

we

-1.48

</strong>

-1.44

"'

-1.43

 oczywiście

-1.41

頷

-1.39

 their

-1.38

 concorda

-1.35

POSITIVE LOGITS

這

1.69

for

1.65

when

1.53

 besar

1.52

 This

1.45

Most

1.45

ेंद

1.44

by

1.43

vaders

1.42

ቍ

1.41

Activations Density 0.018%